Token导航 LogoToken导航

Agent Skill · 研究检索

convert-url-to-nextjs(convert URL TO Next.js 搜索)

用于辅助前端页面、组件、样式和交互逻辑的开发与维护。它适合让 Agent 生成或审查 React、Next.js、Vue、Tailwind、CSS 等相关代码,整理组件结构,或定位布局和性能问题。使用时需要结合项目现有设计系统、路由和构建方式,避免只生成孤立片段;涉及页面改动时,应配合本地预览和构建检查确认视觉效果。

convert-url-to-nextjs 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为操作浏览器;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 66,030。

总安装 269周安装 11GitHub Stars 5下载 87核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:convert-url-to-nextjs(convert URL TO Next.js 搜索)
来源仓库:https://github.com/yigitkonur/skills-by-yigitkonur
仓库路径:skills/convert-url-to-nextjs
安装命令:
npx skills add https://github.com/yigitkonur/skills-by-yigitkonur --skill convert-url-to-nextjs
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/yigitkonur/skills-by-yigitkonur --skill convert-url-to-nextjs

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

convert-url-to-nextjs 将实时网站转换为基于 Next.js App Router 的重建项目。适用于网站复刻、离线保存或跨平台移植任务。强调忠实还原原站结构与样式,而非快速原型或创意设计。建议配合本地预览与构建检查,验证组件层级与交互逻辑正确性。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/convert-url-to-nextjs

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/yigitkonur/skills-by-yigitkonur --skill conver…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要操作浏览器,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

convert-url-to-nextjs 怎么安装?

在终端执行 npx skills add https://github.com/yigitkonur/skills-by-yigitkonur --skill convert-url-to-n…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

convert-url-to-nextjs 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

convert-url-to-nextjs 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

convert-url-to-nextjs 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

convert-url-to-nextjs 用的人多吗?

公开数据是累计安装 269 次,近一周安装 11 次,下载 87 次,GitHub 5 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

convert-url-to-nextjs 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
convert-url-to-nextjs

convert URL TO Next.js 搜索 · 本页

2695操作浏览器未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Capture / Snapshot to Next.js

Convert live sites or saved HTML snapshots into grounded Next.js App Router builds. Treat the source as evidence, not inspiration: capture first when the source is live, extract before build, then prove the rebuilt page matches with visual QA artifacts.

Use this skill when

  • Input is a live URL, a saved snapshot (.html + companion assets), a SingleFile export, or an unpacked offline site capture.
  • The user wants to rebuild, recreate, convert, clone, recover, or extract a site into Next.js.
  • Fidelity matters more than speed or creative redesign.
  • The source repo is partially or fully lost, but production pages or browser artifacts still exist.

Do not use this skill when

  • The only source is a single screenshot, PDF export, Figma design, or loose design idea with no accessible live route or parseable HTML/CSS artifacts.
  • The task is general Next.js UI coding with no reconstruction or grounded extraction requirement.
  • The goal is a redesign, simplification, or same vibe rewrite instead of faithful reconstruction.

Setup

KeyValue
Output locationnextjs-project/ plus .design-soul/ beside the capture or snapshot working root
NamingRoute folders, capture identifiers, manifests, asset folders, CSS classes, and token names use kebab-case. React component files and exported component symbols use PascalCase.
Minified CSS warningCaptured or saved CSS may be minified (no trailing ; before }). All grep patterns in references use [^;}]+ to handle this — do not simplify to [^;]+.

Start with these decisions

SituationAction
Only a live URL existsRun the Capture Wave first. Inventory routes, group page families, choose canonical exemplars, mirror route assets locally, then continue to Wave 0.
Live URL plus a narrow scope requestCapture only the in-scope routes, but still build a route manifest and page-family grouping before implementation.
.html + _files/ folders presentSkip the Capture Wave. Start at Wave 0 and treat the snapshot as the primary artifact set.
.html with adjacent/local .css files but no _files/Use adjacent-asset snapshot mode. Treat the referenced local CSS and assets as the Wave 0 corpus.
.html with inline <style> blocks but no _files/ or local .css filesUse SingleFile fallback mode. Extract from inline styles and document reduced confidence.
package.json + source repo but no live site or snapshotFallback mode only. Read source directly, but keep the same grounding and verification rules.
Request says extract, document, design system, or tokensStop after the Capture Wave or Waves 0–2 unless the user explicitly asks for a build.
Request says rebuild, recreate, convert, clone, recover, or pixel-perfectRun the full pipeline: Capture Wave (if needed) → Waves 0–4.
Source spans many routes or templatesInventory routes first, group them into page families, finish one family completely, then fan out.
Request is ambiguousDefault to grounded extraction first. Do not assume a full rebuild.

Read references/input-output-spec.md for input detection, working-root rules, route normalization, output trees, and ambiguity handling. Read references/capture-workflow.md when the source starts from a live URL.

Operating contract

  1. Capture before extract when the source is live. Build the route inventory, page-family manifest, screenshots, runtime metadata, mirrored asset roots, and grounded DOM artifacts before Wave 0. Read references/capture-workflow.md and references/wave-pipeline.md.
  2. Preserve the strongest offline artifact set. _files/ folders, adjacent local CSS, and inline styles are all valid snapshot evidence. Do not discard a snapshot because it does not match one filename convention.
  3. Route families before route fan-out. Identify the shared shell and canonical page types before rebuilding every route. A convincing recovery comes from reusable structure, not bespoke one-off pages.
  4. Parse, don't guess. Every value must trace to captured CSS, HTML, runtime metadata, or documented behavior. Read references/principles-and-rules.md.
  5. Screenshots are for verification, not invention. Use screenshots and scroll slices to detect missed structure and measure fidelity; do not derive exact token values from pixels when CSS/HTML evidence exists.
  6. Build from extracted values only. tokens.ts, tailwind.config.ts, globals.css, components, and route data must trace to Capture/Wave 0/Wave 1 artifacts. Mark incomplete values UNVERIFIED instead of substituting defaults.
  7. Self-host everything. Fonts, images, icons, and other assets must end up local. No CDN fonts. No remote image URLs.
  8. Write signals last. done.signal and foundation-ready.signal are written only after verification passes.
  9. If something cannot be grounded, mark it UNVERIFIED. Honest gaps are allowed. Invented values are not.

Do this / not that

Do thisNot that
Capture live routes with a browser first and preserve DOM, screenshots, runtime metadata, and mirrored asset rootsJump straight from a live homepage to hand-built JSX
Treat adjacent local CSS and assets as first-class snapshot evidence when _files/ is missingReject a usable snapshot because it does not use the _files/ naming convention
Group routes into page families and choose canonical exemplars before implementationTreat every route as a unique page and duplicate similar layouts
Use CSS Module prefixes, semantic tags, heading outlines, and screenshot coverage together to identify sectionsInfer structure from generic <div> nesting or above-the-fold screenshots alone
Search across the full discovered CSS corpus for the page, then deduplicate shared filesRead minified CSS by eye or treat each hashed CSS file as an isolated system
Carry exact source values through Capture/Wave 0 → Wave 4Round, normalize, or replace values because they look close enough
Run full-page and scroll-slice visual comparison loops before calling a route doneDeclare success from build success or spot-checking only the top viewport
Keep the build dependency set limited to the scaffold rules in references/system-template.mdAdd UI, icon, animation, or font packages for convenience

Required sequence

PhaseGoalRequired referencePrimary outputsGate
Capture WaveRoute inventory, canonical page families, grounded browser artifactsreferences/capture-workflow.md + references/input-output-spec.md + references/wave-pipeline.md.design-soul/capture/route-manifest.json, page-types.md, capture/{route}/dom.html, mirror/, screenshots, runtime-metadata.json.design-soul/capture/done.signal
Wave 0Per-page exploration and deobfuscationreferences/foundations-agent.mdwave0/{page}/exploration.md, deobfuscated.css, behavior-spec.md, assets/wave0/{page}/done.signal
Wave 1Unified design soul by page familyreferences/sections-agent.mdwave1/{group}/design-soul.md, token-values.json, responsive-map.mdwave1/{group}/done.signal
Wave 2Self-contained page build briefsreferences/section-template.mdwave2/{page}/agent-brief.mdwave2/{page}/done.signal
Wave 3Next.js scaffold and design-system foundationreferences/system-template.mdnextjs-project/, wave3/foundation-brief.md, traceability-matrix.mdwave3/foundation-ready.signal
Wave 4Page builds plus visual QA loopsreferences/wave-pipeline.md + references/quality-checklist.mdroute files, page components, copied public assets, visual compare artifactsfinal verification passes

If the user only asked for extraction or documentation, stop after the appropriate phase instead of pushing to build.

Asset and style handling rules

  • Extract CSS custom properties, @font-face, @media, @keyframes, and transition values before touching the build.
  • Treat Wave 1 token-values.json as the source of truth for tokens; Wave 3 only re-expresses those exact values in tokens.ts, Tailwind config, and globals.css.
  • Copy fonts to public/assets/fonts/, images to public/assets/images/..., and icons to public/assets/icons/ or inline them only when the source uses inline SVG.
  • Preserve exact typography, spacing, gradients, shadows, radii, z-index, and breakpoint values.
  • Preserve responsive behavior from real source media queries; do not substitute defaults if the capture uses different breakpoints.
  • Preserve extracted interaction behavior, but only if it can be grounded from CSS or documented JS/runtime behavior specs.
  • For inline style="" attributes or JS-driven states, extract the base CSS and the trigger separately; do not freeze computed runtime values into guessed static styles.

Recovery rules

  • Live site available: capture desktop, tablet, and mobile screenshots plus scroll slices before building. Preserve final DOM, script URLs, runtime metadata, mirrored stylesheets, and exposed asset URLs.
  • Next.js / runtime-heavy site: record __NEXT_DATA__, self.__next_f, build IDs, chunk URLs, and route-level script/style manifests when present.
  • Missing _files/ folder: if the HTML references local CSS files, use adjacent-asset snapshot mode; if it only contains inline CSS, use SingleFile mode; otherwise full reconstruction may be blocked unless a live site can be captured.
  • Missing assets or remote-only assets: download them during extraction and record original → local path mapping.
  • Missing fonts: inspect captured network artifacts and stylesheet URLs. Download font files to public/assets/fonts/ and create @font-face declarations. If the original font cannot be recovered, mark the substitution clearly.
  • External JS (analytics, chat widgets): do not embed third-party scripts blindly. Document them separately and only re-add them if the user explicitly wants that behavior.
  • Missing CSS or JS evidence for a value or behavior: mark it UNVERIFIED and avoid inventing the implementation.
  • Incomplete capture: continue extraction where possible, but do not claim a pixel-perfect rebuild if core layout, type, asset, or below-the-fold evidence is missing.
  • Shared headers, footers, or page-family sections across routes: deduplicate them intentionally; note route-specific overrides rather than re-documenting the whole component each time.

CLI-verifiable acceptance criteria

Every conversion must pass these checks before declaring success:

# Install deps in nextjs-project/ first
npm install

# Type-check passes
npx tsc --noEmit

# Production build succeeds
npm run build

# No UNVERIFIED comments remain in shipped app code
grep -r 'UNVERIFIED' app/ components/ lib/ styles/ && echo "FAIL: unverified values" || echo "PASS"

# No external URLs leak into components or styles
grep -rE 'https?://' components/ styles/ | grep -v '// original:' && echo "FAIL: external URLs" || echo "PASS"

Verification before claiming completion

  • After the Capture Wave: verify route inventory completeness, route normalization, canonical page-family selection, mirrored asset roots, and screenshot coverage at desktop, tablet, mobile, plus scroll slices for long pages.
  • After extraction (before Wave 3): run the grounding test and cross-reference checks from references/quality-checklist.md.
  • Before foundation-ready.signal: verify extracted CSS custom properties and shared tokens are accounted for, allowed dependency set, self-hosted fonts/assets, traceability matrix, and clean TypeScript/build checks.
  • Before final completion: compare source vs build at 1280px, 768px, and 375px using full-page and scroll-segment captures; record visual summaries instead of relying on subjective judgment.
  • If any verification fails, fix it before writing the completion signal or calling the build done.

Reference map

NeedRead
Live-capture prerequisites, working root, route normalization, canonical exemplar selectionreferences/capture-workflow.md
Input detection, capture outputs, output trees, ambiguous requestsreferences/input-output-spec.md
Hard rules: grounding, section identification, zero inventionreferences/principles-and-rules.md
Wave 0 extraction methodreferences/foundations-agent.md
Wave 1 design-soul extractionreferences/sections-agent.md
Wave 2 build-brief formatreferences/section-template.md
Wave 3 scaffold, token wiring, allowed depsreferences/system-template.md
Full orchestration, Capture Wave, and gatesreferences/wave-pipeline.md
Acceptance criteria, visual QA, and failure modesreferences/quality-checklist.md
Section-type and anatomy heuristicsreferences/website-patterns.md

Read only the references needed for the current phase. Keep the top-level skill focused on decisions, sequencing, and guardrails; keep implementation detail in the reference files.

平台分布

Codex

37.4%
按下载量换算33

Claude

33.14%
按下载量换算29

Cursor

17.73%
按下载量换算15

Gemini CLI

8.79%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills